Reconstructing the Human Genetic History of Mainland Southeast Asia: Insights from Genome-Wide Data from Thailand and Laos
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Reconstructing the Human Genetic History of Mainland
Southeast Asia: Insights from Genome-Wide Data from
Thailand and Laos
Wibhu Kutanan *,†,1 Dang Liu †,2 Jatupol Kampuansai,3,4 Metawee Srikummool,5
Downloaded from https://academic.oup.com/mbe/article/38/8/3459/6255759 by Max Planck Institut Fuer Evolutionaere Anthropologie user on 23 September 2021
Suparat Srithawong,1 Rasmi Shoocongdej,6 Sukrit Sangkhano,7 Sukhum Ruangchai,8
Pittayawat Pittayaporn,9 Leonardo Arias 2,10 and Mark Stoneking*,2
1
Department of Biology, Faculty of Science, Khon Kaen University, Khon Kaen, Thailand
2
Department of Evolutionary Genetics, Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany
3
Department of Biology, Faculty of Science, Chiang Mai University, Chiang Mai, Thailand
4
Research Center in Bioresources for Agriculture, Industry and Medicine, Chiang Mai University, Chiang Mai, Thailand
5
Department of Biochemistry, Faculty of Medical Science, Naresuan University, Phitsanulok, Thailand
6
Department of Archaeology, Faculty of Archaeology, Silpakorn University, Bangkok, Thailand
7
School of Public Health, Walailak University, Nakhon Si Thammarat, Thailand
8
Department of Physics, Faculty of Science, Khon Kaen University, Khon Kaen, Thailand
9
Department of Linguistics and Southeast Asian Linguistics Research Unit, Faculty of Arts, Chulalongkorn University, Bangkok, Thailand
10
Centre for Linguistics, Faculty of Humanities, Leiden University, Leiden, The Netherlands
†
These two authors are co-first authors and contributed equally to this work.
*Corresponding authors: E-mails: wibhu@kku.ac.th; stoneking@eva.mpg.de.
Associate editor: Bing Su
Abstract
Thailand and Laos, located in the center of Mainland Southeast Asia (MSEA), harbor diverse ethnolinguistic groups
encompassing all five language families of MSEA: Tai-Kadai (TK), Austroasiatic (AA), Sino-Tibetan (ST), Hmong-Mien
(HM), and Austronesian (AN). Previous genetic studies of Thai/Lao populations have focused almost exclusively on
uniparental markers and there is a paucity of genome-wide studies. We therefore generated genome-wide SNP data for
33 ethnolinguistic groups, belonging to the five MSEA language families from Thailand and Laos, and analyzed these
together with data from modern Asian populations and SEA ancient samples. Overall, we find genetic structure
according to language family, albeit with heterogeneity in the AA-, HM-, and ST-speaking groups, and in the hill tribes,
Article
that reflects both population interactions and genetic drift. For the TK speaking groups, we find localized genetic
structure that is driven by different levels of interaction with other groups in the same geographic region. Several
Thai groups exhibit admixture from South Asia, which we date to 600–1000 years ago, corresponding to a time of
intensive international trade networks that had a major cultural impact on Thailand. An AN group from Southern
Thailand shows both South Asian admixture as well as overall affinities with AA-speaking groups in the region, suggesting
an impact of cultural diffusion. Overall, we provide the first detailed insights into the genetic profiles of Thai/Lao
ethnolinguistic groups, which should be helpful for reconstructing human genetic history in MSEA and selecting pop-
ulations for participation in ongoing whole genome sequence and biomedical studies.
Key words: genome-wide, Mainland Southeast Asia, population interaction, South Asian admixture, cultural diffusion.
Introduction Laos are in the center of MSEA and are characterized by a
Mainland Southeast Asia (MSEA), consisting of Myanmar, diverse landscape involving highlands and lowlands, long
Cambodia, Vietnam, western Malaysia, Laos, and Thailand, coastlines, and many rivers. North-versus-south movements
is a region of enormous diversity, with a population of are facilitated by several rivers, including the Mekong, Chao
263 million people speaking 229 languages belonging to Phraya, and Salaween which are considered to be a key factor
five major language families: Tai-Kadai (TK), Austroasiatic for population movement from southern China and upper
(AA), Sino-Tibetan (ST), Hmong-Mien (HM), and MSEA to lower MSEA. In addition, the Malay Peninsula to the
Austronesian (AN) (Eberhard et al. 2020). Thailand and south acts as a crossroad, facilitating east-versus-west
ß The Author(s) 2021. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License
(http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any
medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com Open Access
Mol. Biol. Evol. 38(8):3459–3477 doi:10.1093/molbev/msab124 Advance Access publication April 27, 2021 3459Kutanan et al. . doi:10.1093/molbev/msab124 MBE
movement by sea and by the narrow width of the Kra (Higham and Thodsarat 2012; Higham 2014). Moreover, the
Isthmus (the narrowest part of the Malay Peninsula). HM- and ST-speaking hill tribes in the mountainous areas of
The geographic heterogeneity of Thailand and Laos is northern Thailand, northern Myanmar, northern Laos, and
reflected in the ethnolinguistic diversity of the region. There southern China migrated to the region during historical times,
are 68.6 million people in Thailand and 6.8 million in Laos, 200 years ago (ya) (Schliesinger 2000; Penth and Forbes
Downloaded from https://academic.oup.com/mbe/article/38/8/3459/6255759 by Max Planck Institut Fuer Evolutionaere Anthropologie user on 23 September 2021
speaking 159 languages belonging to all five major MSEA 2004). Taken together, the archaeological and linguistic evi-
language families (Eberhard et al. 2020). TK languages are dence suggest a complex population structure and history of
widespread in southern China and MSEA and are quite prev- the ethnolinguistic groups of Thailand and Laos.
alent in present-day Thailand and Laos, spoken by 89.4% of This population structure and history remains largely
Thais and 65.7% of Laotians. The major TK speaking groups in unexplored by genetic studies, which have almost exclusively
northern, northeastern, central, and southern Thailand are analyzed autosomal short tandem repeat (STR) loci, and mi-
known as Khonmueang, Lao Isan, Central Thai, and tochondrial DNA (mtDNA), and male specific Y chromosome
Southern Thai or Khon Tai, respectively (Eberhard et al. (MSY) sequences. These studies revealed the relative genetic
2020). AA languages are next in predominance, spoken by heterogeneity of the AA groups and homogeneity of TK
4.0% of Thais and 26.2% of Laotians. In addition, this area is groups (Kutanan et al. 2014, 2017, 2019; Srithawong et al.
also inhabited by historical migrants who speak ST, HM, and 2015, 2020; Kampuansai et al. 2017, 2020) and contrasting
AN languages (frequencies of 3.2%, 0.2%, and 2.8%, respec- male and female genetic histories in the region, especially for
tively, in Thailand; and 2.9%, 4.7%, and 0% in Laos) (Eberhard the matrilocal versus patrilocal hill tribes (Oota et al. 2001;
et al. 2020). The AA, HM, and ST languages are spoken mainly Besaggio et al. 2007; Kutanan et al. 2018a, 2019, 2020). While
by highlanders (the hill tribes) in northern and western genome-wide data provide much richer insights into popu-
Thailand, and in midland and upland regions in Laos, al- lation structure and genetic history, previous genome-wide
though AA languages are also spoken by some lowland studies of Thai/Lao populations are either primarily from
groups, for example, the Mon. AN-speaking groups, such as northern populations (HUGO Pan-Asian SNP Consortium
the Thai Malay (SouthernThai_AN), are distributed in the 2009; Xu et al. 2010; Lipson et al. 2018) or do not provide
Southern Provinces of Thailand, bordering with Malaysia. any information on ethnolinguistic background
Archaeological records document a long history of human (Wangkumhang et al. 2013; Lazaridis et al. 2014). Therefore,
occupation of the area, with modern human remains dated we here generated genome-wide SNP data for 452 individuals
to 46–63 thousand years ago (kya) in northern Laos (Demeter from 33 ethnolinguistic groups from Thailand and Laos, in-
et al. 2012). In addition, cultural remains of SEA hunter- cluding two southern Thai groups that have not been in-
gatherers (e.g., flake stone tools of the Hoabınhian culture) volved in any previous genetic studies, speaking languages
have been found in northern Thailand dating to 35–40 kya that encompass all five language families in MSEA. We ana-
(Shoocondej 2006), and in southern Thailand dating to 27–38 lyzed the allele and haplotype sharing within and between the
kya (Anderson 1990). The transition from a hunter-gatherer Thai/Lao groups and compared them with both modern
tradition to a Neolithic agricultural lifestyle occurs 4 kya all Asian populations and nearby SEA ancient samples. Our
across Thailand and Laos (Higham and Thodsarat 2012; results provide several new insights into the genetic prehis-
Higham 2014); agriculture in MSEA probably has its origins tory of MSEA through the lens of populations from Thailand
in the valley of the Yangtze River in China (Higham and and Laos.
Thodsarat 2012), and ancient DNA evidence indicates that
present-day AA speaking groups in MSEA are most closely
related to Neolithic agricultural communities (McColl et al.
Results and Discussions
2018; Lipson et al. 2018). Overview of Genetic Structure and Allele Sharing
However, the common languages shared by Thais and We generated genome-wide SNP data for 452 individuals
Laotians are TK languages, not AA languages. The origin of from 32 populations from Thailand and 1 population from
the TK languages is thought to be in what is now southern or Laos; when combined with previously published data from 3
southeastern China, and they probably spread to MSEA dur- Thai populations (Lipson et al. 2018; Lazaridis et al. 2014),
ing the Iron Age (Pittayaporn 2014). Whether the spread of there are 482 Thai/Lao samples belonging to 36 populations
TK languages occurred via demic diffusion (an expansion of (fig. 1). We also merged our data with data from modern
people that brought both their genes and their language) or Asian populations generated on the same platform and SEA
cultural diffusion (language spread with at most minor move- ancient samples (supplementary table 1, Supplementary
ment of people) has been debated (Sangvichien 1966; Material online; supplementary fig. 1, Supplementary
Nakbunlung 1994; Pittayaporn 2014). Previous genetic studies Material online). We began with principal component anal-
of uniparental lineages have generally supported demic diffu- ysis (PCA) to investigate the overall population structure of
sion for the maternal side but cultural diffusion from the AA the merged data set and identify any outliers (supplementary
people for the paternal side for major Thai/Lao TK groups fig. 2, Supplementary Material online). After outliers were
(Kutanan et al. 2017, 2018b, 2019). Archaeological evidence removed, PC1 separates South Asian (SA) from East Asian
suggests other population contacts in the region, for example, (EA) groups, with the Kharia (#44), Onge (#45), and Uygur
objects from India that appear during the late Bronze Age and (#65) located in between (fig. 2A; supplementary fig. 3,
Iron Age and involve the AA-speaking Khmer and Mon Supplementary Material online). PC2 separates Northeast
3460Genome-Wide Data of Thai and Lao Populations . doi:10.1093/molbev/msab124 MBE
Downloaded from https://academic.oup.com/mbe/article/38/8/3459/6255759 by Max Planck Institut Fuer Evolutionaere Anthropologie user on 23 September 2021
FIG. 1. Map showing the location of the 36 Thai/Lao ethnolinguistic groups analyzed in this study, color-coded according to language family.
Asian (NEA) groups from SEA groups. With respect to the #34) and SouthernThai_TK (#35), as well as the previously
major MSEA linguistic groups, ST and HM groups are gener- published Thai-HO (#36; this population is from the Human
ally separated from the AA, TK, and AN groups on PC2, while Origins data set of Lazaridis et al. 2014, with no further details
the latter three overlap one another (fig. 2B). Exceptionally, available), Mamanwa (#46) and Cambodian (#51), all show
the Karen speaking ST groups (Karen_ST; #7–9) also overlap additional affinity toward the SA populations (fig. 2A and B).
the AA, TK, and AN groups (fig. 2B), while the ST-speaking Interestingly, the AN-speaking group from Thailand
Lahu from Thailand (#6) and China (#56) and the HM- (SouthernThai_AN; #4), is not close to the AN groups from
speaking IuMien (#3) are grouped with the AA-speaking Taiwan (Amis and Atayal) or Indonesia (Semende and
Kinh (#52) and close to the northern Thai TK groups Borneo; #47-48), but rather they are near the AN-speaking
(N_TK; #21–26). Strikingly, four Thai groups from this study, Negrito group Mamanwa (#46) from the Philippines, and the
that is, the AA-speaking Mon (Monic_AA; #20), AN-speaking Monic_AA, C_TK and S_TK groups. When the PCA was
SouthernThai_AN (#4), and TK-speaking CentralThai (C_TK; performed on only SEA individuals, four poles were observed:
3461Kutanan et al. . doi:10.1093/molbev/msab124 MBE
A 0.06
6464
67 67
64 63
64 63
63
64
B 61
61
61 61
6161
6161
6161
57
59
5959 59 59 59
59
59
64
66
66
66
66 5959 59
67 66
66 66
63
64 58 59 59 59
66 66
6463 64 63
67 57 5858 58
58
5958 59
5959 59
62
63 59 59
65 62 62 67
62
6263
6268
63 57
57
58 5757
57 60 60
62 67 67
67
62
6267 686868
68
68
68
68
68
6868
68
68 58
58 57 57 60 60 59
65 675768 68
68
68 68
68 57 5959
60
65 65 61 61 5959
61
61
61
6161
59 59 60 55 55
59
60 5955
59
65 65
65 61 61
61 59 595959
59 0.025
60555555 59
65 65
65 Northeast Asian 58 5959 60 55
55
60
5455 59
5758 58
5858 59
585959
5959 595954 545455
54
59
57 59 5 54
5459 2
5454
58 57 60660 281
0.03 5857 57 60
60
57
0
59
59
5 5 2131212
2 22
55 15959
59
50 3 11
Downloaded from https://academic.oup.com/mbe/article/38/8/3459/6255759 by Max Planck Institut Fuer Evolutionaere Anthropologie user on 23 September 2021
60 55
59
5555
60
5959 5 5959
159
55
55
55
60 59 55 595959
1 59
60
59 59
55
55
55
60
5455 2 3
54
59 54
54
5 54 54
54
59 221222
28 12 51 56 3
55 2
55 59 1
5913
11
2 20
36 1212
12 56 49 50
49
4950
5 5059
59
1
59
59
311
59
20 12
12 12
566
5249
50
3 5049 49
50
50
49 50
50
49
50
40 39 55 59 1 20 2020 20 20 12 52
56 49
4039 39
39 233 0.000 202020
20 20 20 20 20 20 35
46 46 46 52 56
52 52 3 53 49
52
40 39
403939
40
3939 51 56 20 36 65252
339
9 2020 35 20 46 24 5 56 624 53
PC2
40 12 2020 24 24 24 6
6
PC2
39 39 36 20 35 46 2424 25 622
56
53
6 25
6
3940
40
4040
40
40
4040 20 20 12 5649
12 6 49 49
50
49 20 2020 20 20 2046 34 34 2425
25 252125
22 53
6 3 56 26
53
22
63
65
653
22
53 22
25
26
53 53
40 So 12 56 50 49
50
50
50 4646 46 46
20 35 34 34
46 25 25 25
88 1125 22 325
1526256
256
25
22
262226
26
4040
40 20 12
12 52 5652 350
4949
50
49
49 35
4620 46 20 25 34 11
88810 23
21 6212225
0.00 40 41 uth 20
2020
20 2020
20 20 35 12 4646 5252
46 52
52 353 20 35464646 3446 34 4620
20
2036 20
20 20 24 810910
910 11
878 21 25
25 28
26 2326
2323282827
4141
41
41
4141
41
41414243 As 20
202020202020 36 24
20 4624
24 25622
56
52
5
56
22
56
53 6
24
25
66 3435
36 34 3420 20
36 36 34 51 25 11 7711
723 11
9910 9 23
810
11 23 62328 28
23
27
27 2628 28
41
41 42
4242
42 43 ian 20 20
20 35
20 35
20
20
46 3434 46
24
2424
2524 25 366
53
22
53
25
22
26
625
56
26
53
53
22 34 35 2020 20 11 10 1177 28
921
13
47 134736 2128
947
713 2728
28
27
28
23
42
43
424
4243
42343 20
46 46
4646 35
35 34 25342525
3446 25
2221
825
811 21
2525
36
15
25
26 6
25
653
22
26
26
22
22 35 35 3434
34 5134 7721
34 47
4747 47 28
47 27
43
43
42
42
4343
4343
43
43 46 4646
20
3446
20
46
20
20
36 20
20 2410
11
88
11
10
98 8
21 252325
21
26
2327
26
23
28 35 35 4 4 35 20 34 9 13 13 13 30 28 282827
2035
46
4634 2020 810 25
628 28 20 20 13
43 42 3536
34
36 34
342020
36
20 34 51 25 11
10
11
23
10
1197
711
7
10
10
8923
7
9
2813
28
23
21
23
27
23
27
28
28
26
28
28
27 35 44 4
4 4 36 34 4 51 1047 13
29 7 13
28
32
3535 34 3420 34
3520
34
34
11 7
34921
51
13
47
2113947
747
74736
47
47
13 4727
2823
28
27 4 20 29 34 29 2129 29 29 3232
2732
30
33
28
32 30
30322832
32
44
45
354 4 35
35 2034
35 44 436
202034 5110 13
47 13
13 13
13
28
30 28
28
3227 4
4 44 4 2911 2929 29
292929 29 31 31
28312830
32
32
32
32
32
44 45
4545
45
45 4 4 4
20 2929 29
21
29 29
727
32
28
32
30
3028
32
32
28
33 −0.025 4 29 29 10 28 29 31 33
2828
2831
3132
45
45 45 29 30
32
32
3032
32 51 28
31
4
4444 34
11
29 2929 29
29
29
29 29 31
31
28
31 2832
32 51 33 29
31 30 1414
31 3328
20 44
4444 2910
28 29 28
31
28
33 32 51 29 48 33 48
33
4833
44
44
44
44
4444
44 5151 332931 28
31
30
31
14
33
1428 51 483018
48
4848 48
1414 1414
14
−0.03 44 51 2948 33 48
33
33
48 33
4833 14 27
44 30 14
14
14 51 3328 30
15 51 48 48
48
48 18
33 14 30
33
48 3027
14
51
Eigenvalue
51 30 33
28 18 18 18 16 14
51 18 14 18 1616
10 Southeast Asian 1816
1818 16 181818 16 16
16
18
18
1816 16
16
16 181819 19 1815 16
19
18
18 18 19 19 18
19 19
18
5 1919
118
19158 16
19
17 17
17
17
15
15 1716
1617
17
17
15
17
1517 19
3119
1919 15 17
1717
19
31
19
19 19 15
15 17
17 −0.050 19 15
15 17
15 17
0 18 15 15
15 18 151515
14 14
1 2 3 4 5 6 7 8 9 10 1516 15 16
−0.06 PC
−0.075 −0.050 −0.025 0.000 −0.01 0.00 0.01 0.02
C PC1 PC1
Thai/Lao populations Comparative modern populations Comparative ancient populations
CT
LF
SG
K=5
K=6
H
Hmo
IumongD
So ie gNaw
Lis the---- a------
La u- rnT--3 -------1
Ka --------- i_A ----2
Ka en -------- N--
KarenPw --------- 4
La re Pa o-- --------
LawanSkda ---- -6 -5
Pawa_Waw ng-----
B lau_E est---- ---- ----
Khlangngasteern---- --8 -7
H mu-------- rn----- ----
H inP ------------- ----10 9
M inM ay- -------- ---- ----
Solab al- ---- -------- 12 11
Br --- --- --- --- 14 -1
M --- ------------- ----
Y
Kuhan
Phue----
Shuan-------
Khann-----------
Lu ue 24 --- ------
Blae-- an ----22---2
La ck---- g-- ---- 1
La
Ph n-
N uta ----
Saaw i--- -----
Ka k-- -------- ----
C lue --------------- 29
So lT ---- ---- ---3---3
Th the i--- ---- ----
M ai- rn --- ---- --3
Frbuti---- ha ---- 33 2
Br nc ----- i_T ---3
G min
Lo ati- iwa
Vis i-
M wa ---4 ----- 9
Kh --- rah
O ria- ---- in--
M ge ----- ---- --42
Se anw---- ---- ----
Bome a-- --- --4 3
Am
At is- o--e--- ---- -----
C aya ------------- 46 45
Kinmbl------------------
D h od --- --- --- ----
M -- -- n -- --- 48 7
Shiao--------- --------- ----
La e-- ---- -------- ---- 50 -49
Yi-hu --- ---- --- ---- ----
N ----_C ---- ---- ------5 -51
Haxi- -------- ---- ---------2
Tu
Tujia-
Xib---- ----
H o- --- ---
Oezh---- ---- ----
U roq en ---- ---- ----
D guen- -------- -----60
M ur --- --- --- --- ---
Ja ng---- --------------62-61
H
H Pha
N Gu ae
N ua ha
N
N am
N
N on
N am aiC
N
N ak
BA
IA
IA atK
H
H on
H up aiC
t
t r -
y
en a -- -- --
a
ai- -- ia -- -- -- -4
an -- -- --- -- -5 -5
y
a
o-
o- F
-G aC n
-M C C
-N
-M T
-H
-T H
-T P o
-O
-L aie
i-H
i-S H
i-K u o
uja _
n
Mn
on --- -- -- --1 -1
ala b 1
am --- -- --
o - - -
e
-V
-L o
ah h
u - -- - -
pa ola--- --- --- ---
hu---- h ---
ot Ta --- --- -23
oI
dh ---- ri-
u n ju --
r
a- ri-- ---- ---- --------
on--- --- -------
e -- --
u ha -- -- 1 0
a --- m
rn nd --- ---- 4
-N
h --
am aL Tie
oy
tra ng --- --- ---
--- -- -- -- --- 4 3
on m
an ha av
aiD un
--- --- --- ---- 6 5
ia i- --- ---
sa
in Hu Tie
r T
e
ne -- --- --- --6 --6
m
r
-
-
uiN
an
-
n- --- --
ab ju n
--- --- --- --- ---
H ing n
--- --- --- ---
gL no
-
Ba Ca e
se -------- ---- 4 3
--- --- --- --
aD
- -- --
-
-
- a
--- -- --- 5
a
-
gU
- -
--- --- --- --5 -5
at ng
ap
--- --- --1 --1
-
on u
T
ng
c ve
--- --- --- ---
-
--- --- ---
ie
ag
--- --- --- 6 5
--- --- 8 7
u
- - - -
ju
-
-
--- ---
gR
--- K-
--- --- -6 65
-
u
- -
--- --- ---
-
an
ng
-
--- --- ---
--- ---
- 3
-
--- -35
--- --- 6
-
-
ak
-
-
--- --- ---
-
--- -58 -57
--
C
--- ---
-
-
---
--- ---
-
--- -27 26
av
-
-
---
-
-2 -19
-
- -
---
-
---
68 67
2
---
e
0
---
--4
--3
-2
4
59
4
0
3
6
8
Country (CT) Language family (LF) Subgroup (SG)
Thailand Philippines China Hmong−Mien Central Sudanic Turkic Hmong_HM N_TK
Laos Indonesia Mongolia Austronesian Indo−European Mongolic Karen_ST NE_TK
DR Congo Taiwan Japan Sino−Tibetan Dravidian Japonic Palaungic_AA C_TK
France Cambodia Malaysia Austroasiatic Andamanese Khmu_Katu_AA S_TK
India Vietnam Myanmar Tai−Kadai Tungusic Monic_AA Other
FIG. 2. Population structure analyses. (A) Plot of PC1 versus PC2 for the SNP data for individuals from South Asia, Northeast Asia, and Southeast
Asia. Individuals are numbered according to population, as indicated in supplementary table 1, Supplementary Material online and in the
population labels in panel (C). Thai/Lao groups are colored by language family according to the key at the bottom of panel (C) while other
groups are in gray (see supplementary fig. 3, Supplementary Material online for the same PC plot with all samples colored by country and by
language family). The eigenvalues from PC1 to PC10 are shown on the bottom left side. (B) Plot focusing on Southeast Asian and Chinese
populations speaking AA, AN, HM, ST, and TK languages, zoomed-in from (A). Thai/Lao groups are colored according to subgroup while other
groups are in grey. (C) ADMIXTURE results for K ¼ 5 and K ¼ 6. Each individual is represented by a bar, which is partitioned into K colored
segments that represent the individual’s estimated membership fractions in each of the K ancestry components. Populations are separated by
black lines for modern populations and excavation sites and time periods are separated by black lines for ancient samples. The three colored bars at
the top of the plot indicate the country (top), language family (middle), and subgroup (bottom) for each sample, according to the key at the
bottom. The PCA analysis was performed on the pruned data set of 842 individuals and 153,191 SNPs, while the ADMIXTURE analysis was
performed on the pruned data set of 895 individuals (including 10 Mbuti, 10 French, and 33 ancient individuals) and 158,772 SNPs; the highly
drifted modern populations (Onge, Mlabri, and Mamanwa) and ancient samples were projected in ADMIXTURE analyses (see PCA with ancient
samples projected in supplementary fig. 3, Supplementary Material online).
those groups showing additional affinity to SA populations; Amis and Atayal from Taiwan, and a blue component
the Khmuic/Katuic AA speaking groups (Khmu_Katu_AA); enriched in Khmu_Katu_AA groups from Thailand. Most
the Lahu ST speaking groups; and the HM groups; which of the Thai/Lao TK-speaking groups show two major sources
implies additional admixture or drift has happened in these (black and blue) with the purple component as a minor
groups relative to the other SEA groups supplementary fig. 4, source, except that the C_TK and S_TK groups and Thai-
Supplementary Material online). HO have a substantial fraction of the pink component, as do
We then performed ADMIXTURE analysis to further in- the Monic_AA and Southern Thai_AN. This indication of
vestigate population structure. The lowest cross validation potential relatedness with SA groups is consistent with the
error occurred at K ¼ 5 and K ¼ 6 (supplementary fig. 5, PCA results (fig. 2A and B). At K ¼ 6, there appears a green
Supplementary Material online); corresponding results are component that separates French from South Asian popula-
shown in fig. 2C. For K ¼ 5, there is a brown component tions (fig. 2C). This green component substantially reduces
associated with Mbuti, a pink component appearing in the pink component in the NEA groups but has a negligible
French and Indian groups, a purple component enriched in effect on the SA-related Thai groups. Although increasing K
NEA groups, a black component dominant in AN-speaking values are associated with higher cross-validation errors, the
3462Genome-Wide Data of Thai and Lao Populations . doi:10.1093/molbev/msab124 MBE
additional new components reveal additional population the Khmu_Katu_AA and NortheasternThai_TK (NE_TK)
structure (supplementary fig. 6, Supplementary Material on- groups, but N-Oakaie shares more with the ST-speaking
line). At K ¼ 7, 8, and 9, the Lahu from Thailand and China, Lisu and Lahu groups and HM-speaking Hmong and
the HM-speaking Hmong (Hmong_HM), and Karen_ST IuMien groups. The Iron Age samples show overall less
groups from Thailand are enriched for their own sources, allele-sharing with Thai/Lao groups, whereas the Bronze
Downloaded from https://academic.oup.com/mbe/article/38/8/3459/6255759 by Max Planck Institut Fuer Evolutionaere Anthropologie user on 23 September 2021
respectively. At K ¼ 11, the Soa and Bru (Katuic speaking Age and historical samples from Vietnam and Malaysia
populations of the Khmu_Katu_AA group) stand out with show higher sharing with the Thai/Lao TK and HM groups,
a light brown component. in agreement with the ADMIXTURE results (fig. 2C). Our
To analyze population relationships based on allele shar- results support previous findings (Lipson et al. 2018; McColl
ing, we calculated outgroup f3-statistics of the form f3(X, Y; et al. 2018): the Hoabınhian samples are genetically related to
outgroup) that measure the shared drift between populations Andamanese Onge: the Neolithic samples share ancestry with
X and Y since their divergence from the outgroup (Mbuti). the AA populations (except for the N-Oakaie sample from
Higher outgroup f3 values indicate more shared drift between Myanmar, which shares ancestry with ST speaking popula-
populations. The SouthernThai_AN, Monic_AA, C_TK, and tions); and most Bronze/Iron Age samples are genetically re-
S_TK groups and Thai-HO exhibit the lowest f3-values with lated to both AN and TK speaking populations. However, the
other populations/ancient samples and also with each other inclusion of many more ethnolinguistic groups in our study
(fig. 3), while the HM speaking populations show the stron- brings additional insights, for example, not all AA populations
gest sharing with each other. TK populations exhibit close (Mon and Palaung) are equally related to Neolithic samples,
genetic affinity with each other, except for the C_TK, S_TK, suggesting genetic heterogeneity and the complexity of SEA
and Thai-HO groups, and also share alleles with the HM prehistory.
speaking populations, consistent with results of the Based on the overview provided by the PCA, ADMIXTURE,
ADMIXTURE analysis at K ¼ 8 (supplementary fig. 6, and outgroup f3 results, we focus on the following aspects of
Supplementary Material online). There is higher sharing be- the data: genetic structure and heterogeneity of Austroasiatic
tween the Thai/Lao groups and other SEA and southern speaking groups; genetic structure of the hill tribes; differences
Chinese groups (i.e., TK, HM, and non-NEA ST Chinese among the four major TK speaking groups according to geo-
groups) than with SA and NEA groups (fig. 3). The highest graphic region; and South Asian-related admixture.
sharing was between Thai Lahu and Chinese Lahu. The Amis
and Atayal share more alleles with the TK groups than with Genetic Structure and the Heterogeneity of
the SouthernThai_AN group from Thailand (fig. 3), in agree- Austroasiatic Speaking Groups
ment with ADMIXTURE results (fig. 2C; supplementary fig. 6, AA speakers (comprising 102 million people speaking 167
Supplementary Material online). languages) are widespread across Asia, from South Asia
When ancient samples are included in the analyses of ge- (Bangladesh and India) to southern China and MSEA
netic structure and allele sharing, the two Hoabınhian sam- (Eberhard et al. 2020). There are two competing hypotheses
ples (#69–70) are projected close to the Onge on PCA of AA origins that are related to rice cultivation, namely South
(supplementary fig. 3, Supplementary Material online), while Asian versus Southeast Asian origins (Diffloth 2005; Chaubey
most of the Neolithic samples (#71–79) fall with the AA and et al. 2011); the latter is supported by genetic evidence
AN groups. However, the N-Oakaie sample (#78) from (Chaubey et al. 2011). The AA people in SEA are most likely
Myanmar is closer to ST and HM groups. Most of the related to farmers who cultivated rice and millet and moved
Bronze/Iron Ages samples (#80–82) cluster with the TK from their homeland, probably located near the Yangtze
and AA samples except for the BA-NuiNap samples (#80) River, to the coast and then down the rivers of mainland
from Vietnam, which are close to the Neolithic samples. China to SEA 4 kya (Weber et al. 2010; van Driem 2017;
With respect to ADMIXTURE result at K ¼ 5 (fig. 2C), the Lipson et al. 2018; McColl et al. 2018). However, prior to the
Hoabınhian samples show a major pink component with movement of prehistoric AA-related groups southward,
minor blue and purple components, while all of the present-day MSEA (both upland and lowland) was home
Neolithic samples exhibit a major blue component with mi- to hunter-gatherers whose descendants are genetically re-
nor black, pink, and purple components, except that the lated to groups in southern Thailand and west Malaysia,
purple component is enriched in the N-Oakaie sample such as the Maniq and Jehai (Jinam et al. 2012). The
from Myanmar, and reduced/lacking in the N-GuaChaCave Neolithic farmer expansion did not completely replace the
samples from Malaysia and the N-TamPaLing and N- hunter-gatherers but admixed with some of them, as
TamHang samples from Laos. The purple component is reflected by both ancient and modern DNA studies (Lipson
also enriched in the IA-LongLongRak Iron Age samples et al. 2018; McColl et al. 2018; Kutanan et al. 2017; Liu et al.
from Thailand. The black component is substantially in- 2020).
creased in the Bronze Age and historical samples, such as Previous genetic and linguistic evidence suggested hetero-
the BA-NuiNap and Hi-HonHaiCoTien samples from geneity of the Thai AA people (Xu et al. 2010; Kampuansai
Vietnam and the Hi-SupuHujung and Hi-Kinabatagan sam- et al. 2017; Kutanan et al. 2017; Eberhard et al. 2020) but
ples from Malaysia (a similar pattern is seen in the Thai/Lao further genetic groupings have not yet been investigated.
TK groups). In the outgroup f3 result (fig. 3), the ancient We obtained data for 11 AA speaking populations which
samples N-TamPaLing and N-TamHang share more with can be clustered into four linguistic groups: Monic branch
3463Kutanan et al. . doi:10.1093/molbev/msab124 MBE
CT
LF
Country Language
HmongNjua (CT) family (LF)
HmongDaw Thailand Hmong−Mien
IuMien
SouthernThai_AN
Laos Austronesian
Lisu France Sino−Tibetan
Lahu India Austroasiatic
KarenPwo Philippines Tai−Kadai
Downloaded from https://academic.oup.com/mbe/article/38/8/3459/6255759 by Max Planck Institut Fuer Evolutionaere Anthropologie user on 23 September 2021
KarenPadaung Indonesia Indo−European
KarenSkaw
Taiwan Dravidian
Lawa_Western
Lawa_Eastern
Cambodia Andamanese
Palaung Vietnam Tungusic
Blang China Turkic
Khmu Mongolia Mongolic
HtinPray Japan Japonic
HtinMal
Mlabri
Malaysia
Soa Myanmar
Bru
Mon Subgroup (SG)
Yuan
Khuen Hmong_HM
Phuan Karen_ST
Shan Palaungic_AA
Khonmueang
Khmu_Katu_AA
Lue
BlackTai
Monic_AA
Laotian N_TK
LaoIsan NE_TK
Phutai C_TK
Nyaw S_TK
Saek
Other
Kalueang
CentralThai
SouthernThai_TK
Thai
HmongNjua
HmongDaw
IuMien
SouthernThai_AN
Lisu
Lahu
KarenPwo
KarenPadaung
KarenSkaw
Lawa_Western
Lawa_Eastern
Palaung
Blang
Khmu
HtinPray
HtinMal
Mlabri
Soa
Bru
Mon
Yuan
Khuen
Phuan
Shan
Khonmueang
Lue
BlackTai
Laotian
LaoIsan
Phutai
Nyaw
Saek
Kalueang
CentralThai
Thai
SouthernThai_TK
SG
LF
Thai/Lao groups 0.22 0.24 0.26 0.28 0.3
CT
LF
HmongNjua
HmongDaw
IuMien
SouthernThai_AN
Lisu
Lahu
KarenPwo
KarenPadaung
KarenSkaw
Lawa_Western
Lawa_Eastern
Palaung
Blang
Khmu
HtinPray
HtinMal
Mlabri
Soa
Bru
Mon
Yuan
Khuen
Phuan
Shan
Khonmueang
Lue
BlackTai
Laotian
LaoIsan
Phutai
Nyaw
Saek
Kalueang
CentralThai
SouthernThai_TK
Thai
French
SG
Brahmin_Tiwari
Gujarati
Lodhi
Vishwabrahmin
Mala
Kharia
Onge
Mamanwa
Semende
Borneo
Amis
Atayal
Cambodian
Kinh
Dai
Miao
She
Lahu_C
Yi
Naxi
Han
Tujia
Tu
Xibo
Hezhen
Oroqen
Uygur
Daur
Mongola
Japanese
Ho−PhaFaen
Ho−GuaChaCave
N−GuaChaCave
N−ManBac
N−NamTun
N−MaiDaDieu
N−HonHaiCoTien
N−TamPaLing
N−TamHang
N−Oakaie
N−LoyangUjungCave
BA−NuiNap
IA−VatKomnou
IA−LongLongRak
Hi−HonHaiCoTien
Hi−SupuHujung
Hi−Kinabatagan
LF
Comparative modern groups Comparative ancient groups
FIG. 3. Population allele sharing profiles based on f3 statistics. Heatmap of outgroup f3 statistics (Thai/Lao groups, X; Mbuti) among Thai/Lao
groups (upper) panel, and between Thai/Lao and other comparative modern Asian populations and ancient samples (lower). Black blocks denote
missing values. The two colored bars at the top of the plot indicate the country (top) and language family (bottom) for each comparative
population; and those on the side indicate language family (left) and subgroup (right) for each Thai/Lao group, according to the key at the right.
3464Genome-Wide Data of Thai and Lao Populations . doi:10.1093/molbev/msab124 MBE
CT
LF
HmongNjua
HmongDaw
IuMien
SouthernThai_AN
Lisu
Lahu
KarenPwo
KarenPadaung
>=25
KarenSkaw
Avg. of summed CP length (cM)
Lawa_Western
Lawa_Eastern
Downloaded from https://academic.oup.com/mbe/article/38/8/3459/6255759 by Max Planck Institut Fuer Evolutionaere Anthropologie user on 23 September 2021
Palaung 20
Blang
Khmu
HtinPray
HtinMal 15
Mlabri
Soa
Bru Country Language
Mon
Yuan 10 (CT) family (LF)
Khuen
Thailand Hmong−Mien
Phuan
Shan Laos Austronesian
Khonmueang 5 France Sino−Tibetan
Lue India Austroasiatic
BlackTai Philippines Tai−Kadai
Laotian
LaoIsan
Indonesia Indo−European
Phutai Taiwan Dravidian
Nyaw Cambodia Andamanese
Saek Vietnam Tungusic
Kalueang
CentralThai
China Turkic
A
SouthernThai_TK Mongolia Mongolic
Thai Japan Japonic
HmongNjua Malaysia
HmongDaw Myanmar
IuMien
SouthernThai_AN
Lisu Subgroup (SG)
Lahu Hmong_HM
KarenPwo
Karen_ST
KarenPadaung >=10 Palaungic_AA
KarenSkaw
Lawa_Western Khmu_Katu_AA
Avg. of summed IBD length (cM)
Lawa_Eastern Monic_AA
Palaung
N_TK
Blang
Khmu
7.5 NE_TK
HtinPray C_TK
HtinMal S_TK
Mlabri Other
Soa
Bru
5
Mon
Yuan
Khuen
Phuan
Shan 2.5
Khonmueang
Lue
BlackTai
Laotian
LaoIsan
Phutai
Nyaw
Saek
Kalueang
CentralThai
SouthernThai_TK
Thai B
HmongNjua
HmongDaw
IuMien
SouthernThai_AN
Lisu
Lahu
KarenPwo
KarenPadaung
KarenSkaw
Lawa_Western
Lawa_Eastern
Palaung
Blang
Khmu
HtinPray
HtinMal
Mlabri
Soa
Bru
Mon
Yuan
Khuen
Phuan
Shan
Khonmueang
Lue
BlackTai
Laotian
LaoIsan
Phutai
Nyaw
Saek
Kalueang
CentralThai
Thai
French
Brahmin_Tiwari
Gujarati
Lodhi
Vishwabrahmin
Mala
Kharia
Onge
Mamanwa
Semende
Borneo
Amis
Atayal
Cambodian
Kinh
Dai
Miao
She
Lahu_C
Yi
Naxi
Han
Tujia
Tu
Xibo
Hezhen
Oroqen
Uygur
Daur
Mongola
Japanese
SouthernThai_TK
SG
LF
FIG. 4. Haplotype sharing profiles as inferred by the ChromoPainter and IBD analyses. The color bars at the top denote the countries and language
families while the color bars at the left denote countries and subgroups, according to the keys. (A) Heatmap of ChromoPainter results in which the
recipient Y (Thai/Lao groups) is painted by donor X (Thai/Lao and other modern Asian populations), with Y denoted by each row and X denoted
by each column. The heatmap is scaled by the average length in centimorgans of the summed painted chromosomal chunks of the recipient
individuals from the donor individuals. (B) Heatmap of IBD sharing among Thai/Lao comparisons and between Thai/Lao and other modern Asian
populations. The heatmap is scaled by the average length in centimorgans of summed IBD blocks shared between individuals from the two groups.
Black blocks denote missing values.
(Mon); Khmuic branch (HtinMal, HtinPray, Mlabri, and Supplementary Material online). These revealed some finer
Khmu); Katuic branch (Soa and Bru); and Palaungic branch structure within the AA groups: the Mon_AA group shows
(Lawa_Eastern, Lawa_Western, Palaung and Blang) (Diffloth excess sharing with Indian donors (discussed in more detail
2005; Sidwell 2014). However, based on the PCA (fig. 2B), the below); Khmu_Katu_AA groups show strong intragroup
Thai AA speaking groups can be roughly divided into three sharing but less sharing with other groups except for between
groups: Palaungic_AA (Lawa_Western, Lawa_Eastern, the Soa and most NE_TK groups; and Palaungic_AA groups
Palaung and Blang; #10–13); Khmu_Katu_AA (Khmu, show various sharing patterns, for example, a broad sharing
HtinPray, HtinMal, Mlabri, Soa and Bru; #14–19); and profile of the Blang with several other groups versus strong
Monic_AA (Mon; #20). The ADMIXTURE results at K ¼ 5 self-painting only of the Palaung, and strong sharing among
also indicated that the AA-speaking groups can be clustered the Lawa_Eastern, Lawa_Western, Karen_ST groups, and TK-
into three groups: the Palaungic_AA group exhibits two ma- speaking Shan.
jor sources (blue and purple) with the black component as a We next computed f4-statistics of the form f4(group 1,
minor source; the Monic_AA group possesses the pink com- group 2; group 3, Mbuti), where group 1 and group 2 are
ponent; and the Khmu_Katu_AA group has a reduced fre- different AA groups while group 3 is from a different language
quency of the purple component (fig. 2C). family/subgroup. By convention, a Z-score > 3 or < 3
To further investigate the genetic structure of AA indicates that group 3 shares significant excess ancestry
groups, we carried out haplotype-based analyses, namely with group 1 or 2, respectively; nonsignificant Z-scores indi-
ChromoPainter and sharing of segments that are identical cate that groups 1 and 2 form a clade and share equivalent
by descent (IBD) (fig. 4; supplementary figs. 7–9, amounts of ancestry with group 3. The AA groups show a
3465Kutanan et al. . doi:10.1093/molbev/msab124 MBE
very heterogenous profile, for example, when compared to includes several admixture events, and indicates that the
the AA-Palaung, most of the other AA groups have additional Khmu_Katu_AA and Palaungic_AA subgroups are more
affinity to TK and AN groups (and some with the ST groups), closely-related, while the Monic_AA subgroup is distin-
whereas the Palaung shows excess sharing with all the other guished from these by N-Indian-related ancestry, in agree-
groups compared to the Mon (supplementary fig. 10A, ment with the results of other analyses (figs. 2 and 4A).
Downloaded from https://academic.oup.com/mbe/article/38/8/3459/6255759 by Max Planck Institut Fuer Evolutionaere Anthropologie user on 23 September 2021
Supplementary Material online; supplementary table 2, Overall, the genetic evidence indicates that the Thai AA
Supplementary Material online). Consistent with the haplo- speaking populations fall into 3 primary groups: Monic_AA,
type sharing profile of the Palaungic_AA group, the Blang has Khmu_Katu_AA and Palaungic_AA (figs. 2–4; supplemen-
a broader excess sharing with all other subgroups except for tary fig. 12, Supplementary Material online). The language
the TK-speaking Khuen, the ST-speaking Lisu, and the HM of Mon is in the Monic branch, the sister clade of Aslian
groups, and the Lawa groups seem to have additional affinity and Nicobarese, while the linguistic branch of
to the Khmu_Katu_AA and the Karen_ST groups (supple- Khmu_Katu_AA groups are Khmuic for HtinMal, HtinPray,
mentary fig. 11A, Supplementary Material online; supplemen- Mlabri, and Khmu, and Katuic for Soa and Bru; the Palaungic
tary table 3, Supplementary Material online). However, within branch includes languages of the Lawa_Eastern,
the Khmu_Katu_AA group, the Khmuic branch groups tend Lawa_Western, Palaung, and Blang. In contrast to linguistic
to show excess sharing with the Palaungic_AA and the ST studies placing Khmuic and Palaungic languages in the same
groups compared to the Katuic branch groups (supplemen- clade (Diffloth 2005), we find a closer relationship between
tary fig. 11B, Supplementary Material online; supplementary populations who speak Khmuic and Katuic, which might be
table 3, Supplementary Material online). explained by the concept of center of gravity (Blench 2015).
We further investigated the groupings among AA Thai/ This idea proposes that after the Neolithic expansion of AA
Lao groups by f4-statistics of the form f4(East Asian group, ancestors from southern China to MSEA, early AA speakers
Han Chinese; AA Thai/Lao group, Mbuti), to see if any of the were concentrated along the middle Mekong in present-day
AA groups showed different affinities with any East Asian northern Laos. Some groups subsequently moved westward
groups in comparison with Han Chinese (supplementary ta- and were the ancestors of Palaungic and Monic groups, and
ble 3, Supplementary Material online). Based on the allele and during this process they came into contact with different
haplotype sharing profiles (figs. 3 and 4), we used Atayal, Dai, linguistic groups (e.g. Mon with Burmese ancestors,
Cambodian, Miao, and Naxi as representative East Asian Lawa_Eastern and Lawa_Western with Karen_ST, and
groups speaking AN, TK, AA, HM, and ST languages, respec- Palaung with ST groups from NEA), as shown by population
tively. The grouping among AA Thai/Lao groups was also structure and relationship analyses and f4 tests (figs. 2–4;
supported by this test; the Monic_AA show excess sharing supplementary fig. 11, Supplementary Material online; sup-
only with the Dai, while the Khmu_Katu_AA and plementary table 3, Supplementary Material online). These
Palaungic_AA groups are distinguished by the former sharing different contact histories would promote subsequent differ-
excess ancestry with Atayal and having no significant Z-scores entiation of the Palaungic and Monic groups from their
with Cambodian versus Han, while the latter have no signif- Khmuic and Katuic ancestors. Meanwhile, the Khmuic and
icant Z-scores with Atayal and share excess ancestry with Han Katuic ancestors might have moved up and down the
when compared with Cambodian. These results suggest more Mekong and had more contact with each other, thus ac-
AN/TK and AA related ancestry in the Khmu_Katu_AA counting for their closer genetic relationship with each other.
group, and more Han related ancestry in the Palaungic_AA In this region, the Khmuic and Katuic speaking people may
group. have also interacted with TK groups in Laos and Northeastern
We finally built admixture graphs using AdmixtureBayes, Thailand, promoting their genetic affinity (figs. 2B, 3, and 4;
and then further investigated these admixture graphs with supplementary table 3, Supplementary Material online).
qpGraph. To begin with, we built a backbone admixture However, some differentiation between the Khmuic and
graph with the outgroup Mbuti, N_Indian, and the following Katuic groups can be seen in the haplotype sharing (fig. 4)
representative East Asian groups: AA-speaking Cambodian, and ADMIXTURE results for K ¼ 10 (supplementary fig. 6,
AN-speaking Atayal, TK-speaking Dai, HM-speaking Miao, Supplementary Material online). Additional studies of AA
and ST-speaking Naxi (fig. 5A). Another f4 test with Amis, groups from Thailand (e.g. Pearic and Khmer speaking
She, and Yi as alternative AN, HM, and ST representative groups) and other MSEA countries are needed to provide
groups, respectively, was performed to verify that our choice more insights into the genetic structure of AA-speaking
of representative groups is not biased in distinguishing the people.
fine-scale relationships within each language family (supple-
mentary fig. 13, Supplementary Material online). In the back- Genetic Structure of the Hill Tribes
bone graph, the first split separates the N_Indian from the Consisting of 700,000 people, there are nine officially rec-
East Asian groups, then the Naxi are separated from the other ognized hill tribes in Thailand: the AA-speaking Lawa
groups. The ancestor of Atayal and Dai is admixed from (Lawa_Eastern and Lawa_Western), Htin (HtinMal and
ancestors of N_Indian and Miao with 6% and 94% ancestry, HtinPray) and Khmu; the HM-speaking Hmong
respectively. The ancestor of Cambodian is admixed with 73% (HmongNjua and HmongDaw) and IuMien; and the ST-
ancestry from the ancestor of Dai and 27% from the ancestor speaking Karen (KarenPwo, KarenPadaung, and
of all East Asian groups. The graph of AA groups (fig. 5B) KarenSkaw), Lahu, Lisu, and Akha (Schliesinger 2000, 2001;
3466Genome-Wide Data of Thai and Lao Populations . doi:10.1093/molbev/msab124 MBE
A r B r C r
31 31 30 30 31 31
Mbuti Mbuti Mbuti
2 7 8 3 9
Downloaded from https://academic.oup.com/mbe/article/38/8/3459/6255759 by Max Planck Institut Fuer Evolutionaere Anthropologie user on 23 September 2021
N_Indian
1 10 10 18
9
N_Indian
2 ~0 72%
28% 5%
1 2 1
6%
Naxi ~0 1 95%
0.4
94% 0.4 Naxi 5 1
Miao 27% 21% 0.5 1 13% Miao
71%
2 0.5 1
Paluangic_AA 29%
2
79% 0.5 Hmong_HM Dai
12 0.3
~0
Atayal 0.1 19% 81% 87%
10% 90%
73% IuMien
0.3
N_Indian
Dai
0.3 ~0 2
1
Monic_AA Cambodian Khmu_Katu_AA
Cambodian
Austroasiatic
Austronesian
Hmong−Mien
Sino−Tibetan
D E F Tai−Kadai
r r r
31 31 28 28 31 31
Mbuti Mbuti
Mbuti
1 14 1 7
10 11
2 10
2 2
2 9
37% N_Indian
N_Indian
63% 0.1 0.1 27% 1 0.3
1 1
1
12% NE_TK 73% 0.5 5
2 0.5 1 27%
17% 2 ~0 88% Dai
13% Lisu Naxi 90%
N_Indian ~0 7
6 4 10%
83% 0.3 0.2 Cambodian Atayal
Lahu 71%
42%
N_TK C_TK 73%
1
58%
0.1
~0
87% 74% 26% 29%
S_TK SouthernThai_AN
0.3 2 2
Cambodian Karen_ST Miao
FIG. 5. Admixture graphs for the Thai/Lao groups, for each language family. The node r denotes the root. White nodes denote backbone
populations. Backbone population labels and Thai/Lao nodes are colored according to language family. Dashed arrows represent admixture
edges, while solid arrows are drift edges reported in units of FST 1,000. (A) backbone populations (worst-fitting Z ¼ 0.861). (B) AA groups (worst-
fitting Z ¼ 2.101). (C) HM groups (worst-fitting Z ¼ 2.028). (D) ST groups (worst-fitting Z ¼ 2.873). (E) TK groups (worst-fitting Z ¼ 2.270).
(F) AN group (worst-fitting Z ¼ 1.713).
3467Kutanan et al. . doi:10.1093/molbev/msab124 MBE
Penth and Forbes 2004). Living in a remote and isolated re- contrast, the Karen in Thailand are refugees who claim to
gion of Thailand, the hill tribes are of interest for their cultural be the first settlers in Myanmar before the arrival of Mon and
variation in postmarital residence patterns, that is, patrilocal- Burmese people, and moved from Myanmar beginning
ity versus matrilocality (Oota et al. 2001; Besaggio et al. 2007; around 1750 A.D. due to the growing influence of the
Kutanan et al. 2019, 2020). Most of the hill tribes are isolated Burmese (Kuroiwa and Verkuyten 2008; Gravers 2012). The
Downloaded from https://academic.oup.com/mbe/article/38/8/3459/6255759 by Max Planck Institut Fuer Evolutionaere Anthropologie user on 23 September 2021
from the lowlanders and from each other, which enhances Lawa share ancestry with the Karen_ST (fig. 4; supplementary
genetic drift and inbreeding, as found in previous studies of fig. 5, Supplementary Material online), in agreement with
autosomal STR (Kampuansai et al. 2017) and mtDNA and previous findings of shared MSY haplotypes (Kutanan et al.
MSY variation (Kutanan et al. 2020). 2020). Genetic relatedness between Karen and Lawa groups
Here, we investigated eight of the official hill tribes (all but was also reported in a previous genome wide study (Xu et al.
the Akha) and the Mlabri, who are not officially regarded as a 2010). In northern Thailand, Lawa and Karen had been in
hill tribe but live in the mountainous area. All of them exhibit contact with one another since around the 13th century
high within-population IBD sharing (supplementary fig. 8, A.D., during the Lanna Period (Lewis and Lewis 1984).
Supplementary Material online), as expected given the results Because the languages of AA-speaking Lawa and ST-
of previous studies that suggested high levels of isolation and speaking Karen are different, geographic proximity along
strong genetic drift. The Mlabri in particular show the greatest the border between northern/northwestern Thailand and
levels by far of within-group IBD sharing, in agreement with Myanmar is the most likely factor that promoted admixture
their enhanced self-painting in the ChromoPainter analysis between these groups.
(fig. 4A). In the ADMIXTURE results at K ¼ 10, four groups The IuMien and Hmong are descended from proto-HM
stand out with their own ancestry components (supplemen- groups from central and southern China (Wen et al. 2005)
tary fig. 6, Supplementary Material online): Lahu (light green), and are linguistically related; there is no significant sharing of
Karen_ST (gray), Htin (Mal and Pray), Khmu (mint), and ancestry between HM and non-HM groups in the f4 analyses
Hmong_HM (peach). In contrast, the Lawa (Eastern and (supplementary fig. 10B, Supplementary Material online; sup-
Western), IuMien, and Lisu do not stand out in the plementary table 2, Supplementary Material online).
ADMIXTURE analysis, and they have relatively less within However, they still behave differently in many analyses
group IBD sharing compared to other hill tribes (supplemen- (figs. 3–5; supplementary figs. 6 and 12, Supplementary
tary fig. 8, Supplementary Material online). This was further Material online). The Hmong show genetic signatures of iso-
revealed by excess allelic sharing with many other populations lation, such as higher IBD sharing within groups (supplemen-
in the f4 results (supplementary tables 2 and 3, tary fig. 8, Supplementary Material online), in agreement with
Supplementary Material online) and haplotype sharing with a previous study of uniparental markers (Kutanan et al. 2020),
other groups (fig. 4A; supplementary fig. 7, Supplementary whereas the IuMien show affinities not only with the Hmong
Material online). but also with TK speaking groups and ST speaking Lahu from
The Lawa belong to the Palaungic_AA group, which in the both Thailand and China (fig. 4). The differential affinities of
admixture graph for AA groups receives ancestry from the HM groups to TK and ST groups has also been shown in two
ST-Naxi (fig. 5B). We further built admixture graphs for the recent genome-wide studies (Liu et al 2020; Xia et al. 2019). In
HM and ST hill tribes. For the HM groups (fig. 5C), there is a addition, the sharing of features between IuMien (but not
divergence between the Dai and a Miao-Hmong clade, while Hmong_HM) and Sinitic languages (Blench 2008) indicates
the IuMien are admixed with 29% ancestry from an ancestor that IuMien similarities with other East Asian populations is
of the Hmong and 71% from an ancestor of the Dai. The evident both genetically and linguistically. The higher genetic
additional TK-related ancestry in IuMien is consistent with isolation of the Hmong could reflect cultural isolation arising
haplotype-sharing and f4 results (fig. 4; supplementary fig. 12, from a strong preference for marriage within Hmong groups,
Supplementary Material online). The graph of ST groups while the lower genetic isolation of the IuMien could reflect
indicates that Lisu, Lahu and Naxi form a clade, while the the pronounced IuMien cultural preference for adoption
Karen_ST have additional Cambodian-related ancestry (Schliesinger 2000; Jonsson 2005; Besaggio et al. 2007).
(fig. 5D); this AA-related admixture in the Karen is in agree- The Lisu and the Lahu are originally from southern China
ment with the haplotype-sharing results (fig. 4), and the di- and speak closely related languages that belong to the Loloish
vision of Lahu/Lisu versus Karen_ST groups is also supported branch of ST (Bradley 1997). Shared genetic ancestry between
by f4 results (supplementary fig. 10C, Supplementary Material Lisu and Lahu is evident in the haplotype sharing and admix-
online). ture graph results (figs. 4 and 5D), although there are differ-
These results indicate that not all hill tribes can be char- ences: Lisu have mixed ancestries probably due to Sinicization
acterized simply by high degrees of isolation and genetic drift; in southern China before movement to Thailand (Schliesinger
the Lawa, IuMien, and Lisu instead seem to have had more 2000) or interactions with northern Thai lowlanders after
interactions with other groups, and so we will focus further settlement in Thailand (Penth and Forbes 2004), while the
discussion on these three hill tribes. The Lawa (Eastern and Lahu are more isolated, e.g. the ADMIXTURE result for K ¼ 7
Western) are the native groups of northern Thailand and (supplementary fig. 6, Supplementary Material online) and
inhabited lowland areas before some of them moved to the the IBD sharing results (supplementary fig. 8,
highlands (Lawa_Western) while others remained in the low- Supplementary Material online), in agreement with a previ-
lands or mid-lands (Lawa_Eastern) (Nahhas 2007). By ous study of uniparental markers (Kutanan et al. 2020). There
3468Genome-Wide Data of Thai and Lao Populations . doi:10.1093/molbev/msab124 MBE
is strong ancestry sharing between the Thai Lahu and Chinese groups that expanded from China. A previous genome-
Lahu (figs. 3 and 4), and the Chinese Lahu are moreover wide study also reported substructure of Thais in each region
genetically similar to Vietnamese Lahu (Liu et al. 2020), indi- (Wangkumhang 2013); however, these previous studies did
cating a close relationship among Lahu from MSEA and not investigate this substructure in detail.
China. In this study, we investigated one TK population from Laos
Downloaded from https://academic.oup.com/mbe/article/38/8/3459/6255759 by Max Planck Institut Fuer Evolutionaere Anthropologie user on 23 September 2021
Finally, though the Mlabri are not officially regarded as a and 15 TK populations from Thailand that can be grouped by
hill tribe, this minority group is of interest due to their unique geographic region: northern Thailand (N_TK), northeastern
hunting-gathering life style, enigmatic origin, and very small Thailand (NE_TK), Central Thailand (C_TK), and Southern
census size (400 individuals) (Eberhard et al. 2020). The Thailand (S_TK). Based on the PCA (fig. 2B), the TK groups
Mlabri language belongs to the Khmuic branch of AA lan- from different geographic regions in Thailand show different
guages that is also spoken by their neighbors, Htin (Mal and relationships; the N_TK groups are close to the Palaungic_AA
Pray subgroups) and Khmu, suggesting shared common an- groups, AA-speaking Kinh, AN groups from Taiwan (#49–50)
cestry, and oral tradition indicates that the Htin are the and the Philippines (#46), while the northeastern Thai TK
ancestors of the Mlabri (Oota et al. 2005). A previous groups (NE_TK; Black Tai, Lao Isan, Phutai, Nyaw, Saek, and
genome-wide study also supported genetic affinities between Kalueang; #27 and #29-33) are close to the Khmu_Katu_AA
the Mlabri and the HtinMal (Xu et al. 2010), while uniparental groups. The TK speaking Laotian (#28) are grouped with the
studies indicate paternal relationships among Mlabri, NE_TK groups. The central and southern Thai TK groups
HtinMal, HtinPray, and Khmu and an oral tradition, versus (C_TK and S_TK; CentralThai and SouthernThai_TK; #34
maternal genetic relationships among Mlabri and Katuic- and #35) and Thai-HO (#36) are close to the Monic_AA
speaking Soa and Bru from northeastern Thailand (Kutanan groups. In accordance with the PCA results, ADMIXTURE
et al. 2018a). Our present results also support genetic relat- results at K ¼ 11 also show the different TK-speaking groups
edness among Mlabri, Htin (Mal and Pray), Khmu, Soa, and can be distinguished: the blue component is now enriched
Bru within the Khmu_Katu_AA group (fig. 2B; supplemen- mostly in the N_TK group, the additional light brown com-
tary figs. 6 and 7, Supplementary Material online). The Mlabri, ponent is enriched in the NE_TK group, and the C_TK and
Htin, Khmu, Soa, and Bru all migrated from Laos about 100– S_TK group possess the additional pink component as men-
200 years ago (Schliesinger 2000), thus close relatedness tioned previously (supplementary fig. 6, Supplementary
among them might reflect gene flow among various groups Material online).
in Laos before their independent migrations to Thailand. Some finer structure within the Thai TK groups is revealed
However, the Mlabri stand out among these groups in exhib- by ChromoPainter analysis (fig. 4A; supplementary fig. 7,
iting extremely high levels of within-group IBD sharing (sup- Supplementary Material online): N_TK populations show
plementary fig. 8, Supplementary Material online), indicating strong sharing with each other and the Dai, though the
strong genetic drift and isolation, consistent with previous Shan show additional sharing with the Lawa_Eastern and
investigations of mtDNA, Y chromosome, and autosomal Karen_ST groups. The NE_TK groups show strong sharing
diversity (Oota et al. 2005; Xu et al. 2010; Kutanan et al. with the Khmu_Katu_AA group, Cambodian, Borneo, and
2018a). Both the small census size and recent origin within Dai. Notably, the Laotian show a relatively broader sharing
the past 1,000 years (Oota et al. 2005), combined with geo- profile and high sharing with the HM groups, whereas the
graphic isolation, could account for the very low genetic di- BlackTai show a strong selfpainting profile. In addition to
versity of this group. strong sharing with Khmu_Katu_AA groups, the C_TK group
shows an excess sharing with the Indian donors, which is
Differences among the Four Major TK Speaking similar to the profile of Thai-HO. The S_TK group also shows
Groups According to Geographic Region a similar profile as C_TK but additional sharing with the AN-
With an origin from south/southeastern China (Sun et al. speaking Mamanwa, Borneo, and Semende, which is similar
2013; Pittayaporn 2014), the TK language family comprises to the profile of the SouthernThai_AN (who show even
around 95 languages spoken by 80 million people in north- stronger and broader sharing with the other AN groups).
east India, southern China, Vietnam, Myanmar, Cambodia, The results of f4-statistics of the form f4(TK group 1, TK
Thailand, and Laos (Eberhard et al. 2020). A common origin group 2; non-TK group 3, Mbuti) show that, in particular, the
of TK and AN language families in southern China was sug- profiles of NE_TK and N_TK groups show strong excess shar-
gested previously based on linguistic and genetic evidence ing with each other and the HM groups, followed by ST and
(Thurgood 1994; Sagart 2004; Kutanan et al. 2018b; Yang AA groups (supplementary fig. 11C–E, Supplementary
et al. 2020). The TK languages spread to MSEA around 1–2 Material online; supplementary table 3, Supplementary
kya (Pittayaporn 2014), and previous genetic studies esti- Material online). Many of the highest Z-scores come from
mated an expansion time for TK groups 2 kya (Kutanan comparisons involving the Laotian population (supplemen-
et al. 2019) and found relatedness between modern TK pop- tary figs. 10D and 11C, Supplementary Material online; sup-
ulations and ancient Iron Age samples (McColl et al 2018). plementary tables 2 and 3, Supplementary Material online), in
MtDNA and MSY data indicate contrasting genetic variation agreement with their broader haplotype sharing profiles
and genetic differences between major TK groups in the (fig. 4). In addition, we found that Thai-HO and
North, Northeast, and Central regions of Thailand (Kutanan CentralThai form a clade in all the tests (Z scores within 6
et al. 2019), suggesting different migration routes of TK 1.5), suggesting their close relationship in agreement with
3469You can also read
